What I found interesting this time

“Firefox Is Now the Last Major Browser That Still Supports UBlock Origin,” n.d. https://www.pcworld.com/article/3212428/firefox-is-now-the-last-major-browser-that-still-supports-ublock-origin.html.

“Separating AI’s Technological Problems from Its Capitalism Problems,” n.d. https://www.schneier.com/blog/archives/2026/08/separating-ais-technological-problems-from-its-capitalism-problems.html

Different take on problems which are caused by high paced AI development. This time it’s "fear of capitalism collapse".
I see the points of the author. The idea that we could sustainably created AI is beautifull. Sadly it’s not possbible in modern world.
Claim that we should separate AI technological problems with capitalism problems is wrongly put because we live under capitalism. You can’t separate it.

“There Are No Lossless Transformations of Natural-Language Text,” n.d. https://simonwillison.net/2026/Aug/11/there-are-no-lossless-transformations-of-natural-language-text/

Valid point about LLMs which I think we all miss everyday. Nowdays management requires us to speed up and deliver faster with usage of text generators but in the and we bare the responsibility on the quality of what’s generated.
"There are no lossless transformations of natural-language text" should make you think before you simply apply LLMs on your Docs.

“Compression Is Prediction,” n.d. https://ngrok.com/blog/compression-is-prediction

Blog post where author compares one of compression algorithms to LLMs. #compression #math #llms

“You Have Been Mislead about Lightbulbs,” n.d. https://maurycyz.com/misc/tungsten/

Why am I even reading about lightbulbs XD? So yeah. Apparently its hard to make long lasting lightbulbs which would produce a lot of light and not consume decent amount of energy.

“No, Local Models Will Not Win,” n.d. https://www.seangoedecke.com/local-models-will-not-win

Author shares his opinion on why "local models" will never be the solution for people to use for problem solving tasks.
I think it’s safe to say that we don’t know how the future will look like. Personally I place myself within spectrum of "someone will optimize LLMs to a point where you will be able to run frontier models on your phone" as there is really not that much going on there. We just need proper optimizations and innovative ideas.

“What Casey Newton Misunderstands about Powerful AI,” n.d. https://www.jonathanychan.com/blog/what-casey-newton-misunderstands-about-powerful-ai/.

“Regressive JPEGs:” n.d. https://maurycyz.com/projects/bad_jpeg/

Interesting blog post where author manipulated JPEG headers to lazy load images. The final demo with animations containing 90 frames is amazing!